TMP0203: CPU temperature is greater than the upper critical threshold

Dell EMC PowerEdge Servers with iDRAC9 may report CPU x temperature is greater than the upper critical threshold events when CPU workload spikes at or near CPU Power Max. When transient power bursts occur to meet additional CPU demand, the processor temperature may briefly exceed upper critical threshold. For example, a server that is operating at 50-60% utilization and spikes to 100% utilization for 5-20 seconds may exceed upper critical threshold for CPU temperature briefly. When this threshold is exceeded, events will be recorded in the System Event Log and Lifecycle Log. When the peak transient spike is over and CPU temperature returns to normal, an event is recorded indicating CPU x temperature is within range.

When transient spikes like this occur, the two events will typically occur within 5-20 seconds of each other. Please refer to examples below as guidance for these types of transient events.

Lifecycle Log:

When these temperature thresholds are met, Intel processors may throttle to reduce power consumption and lower CPU temperature.

While these types of transient CPU performance spikes are not abnormal, iDRAC9 Engineering continues to fine tune the thermal algorithm to prevent these events. For example, iDRAC9 4.22.00.00 and iDRAC9 4.40.00.00 will include thermal improvements specific for this sequence. To ensure that the latest dynamic thermal algorithms are installed on Dell EMC PowerEdge servers, update to the latest available iDRAC9 firmware.

Workarounds:

End-users can manually modify the system thermals to prevent these transient events from spiking CPU temperature. Use either of the following workarounds to increase the fan speed baseline and maintain lower CPU temperature.

Max Performance Profile

System Thermal Profile Optimization can be modified to Maximum Performance (Performance Optimized). This thermal profile carries the following advantages:

• Reduced probability of memory or CPU throttling.
• Increased probability of turbo mode activation.
• Generally, higher fan speeds at idle and stress loads.

Thermal Profile Optimization can be modified through following methods:

iDRAC9 GUI >> Configuration >> System Settings >> Hardware Settings >> Cooling Configuration

racadm set System.ThermalSettings.ThermalProfile

racadm>>racadm set System.ThermalSettings.ThermalProfile 1

[Key=System.Embedded.1#ThermalSettings.1]

Object value modified successfully

Supported Values:

0 - Default Thermal Profile Settings

1 - Maximum Performance

2 - Minimum Power

3 - Sound Cap

Fan Speed Offset

Fan speed offset allows you to increase the system fan speed with four incremental steps. These steps are equally divided between the typical baseline speed and the maximum speed of the server system fans. A fan speed offset causes fan speeds to increase (by the offset % value) over baseline fan speeds calculated by the Thermal Control algorithm. Possible values are:

• Low Fan Speed — Drives fan speeds to a moderate fan speed.
• Medium Fan Speed — Drives fan speeds close to medium.
• High Fan Speed — Drives fan speeds close to full speed.
• Max Fan Speed — Drives fan speeds to full speed.
• Off — Fan speed offset is set to off. This is the default value. When set to off, the percentage does not display. The default fan speed is applied with no offset. Conversely, the maximum setting will result in all fans running at maximum speed.

Fan Speed Offset can be modified through following methods:

iDRAC9 GUI >> Configuration >> System Settings >> Hardware Settings >> Cooling Configuration

racadm set System.ThermalSettings.FanSpeedOffset

racadm>>racadm set System.ThermalSettings.FanSpeedOffset 2

[Key=System.Embedded.1#ThermalSettings.1]

Object value modified successfully

Supported Values:

0 - Low

1 - High

2 - Medium

3 - Max

255 - Off